Master Clock for your EtherREGEN

[Clockmeister]

6 hours ago, LowMidHigh said:

 

I'm a lowly layman, who is guided by John"s views on such matters. 

. That topic was discussed when a proposal of equipping the ER with said toggle was floated up and   summarily rejected. I suppose Mutec offers two batteries of BNC ports, 50ohn and 75ohm respectively, for that reason. 

 

This point has been mooted about in many circles here in the UK impedence matcing is critical for accurate signal fidelity whatever area of electronics you may happen to work in. In audio design it is rarely taken into account, however though John S has taken the time to make the effort to account for these criteria.

 

In my own lab I use a Mutec Ref 10 SE-120 as a master 10Mhz reference  clock for the test equipment especially useful when performing multi domain analysis in both the time and frequency domains simultaneously. In essence using both the high specification digital oscilloscope and real time spectrum analyser together to see a wider range of signal infidelities happening instantaneously it's a very useful ability to have when looking at hard to find fast moving issues with circuits and rf receiver applications.

 

The Mutec also does make for a rather good audio clock as well as ones recides in the main system.  although all of the audio word clock connected turns are genuine 50 ohms which is a far more robust and desirable impedence pathway, however finding the correct matching terminations may not be do easy

 

 

 

 

 

[MartinT]

On the back of my success with the BG7TBL, and @zerung going for the Adjust version with all squarewave outputs, and @Clockmeister comments on impedance, I have ordered an Adjust version with 75 ohm impedance throughout, 10MHz on all outputs and a DAPU clock.  He is building it for me.

 

Well, why not?  They are inexpensive enough that you can play with them.  I might get quite serious with this one, put some trick electrolytics and bypass ceramics in, get it running as quiet as possible.

[AnAndGate]

On 6/27/2020 at 9:41 PM, Clockmeister said:

I am sure Martin T will be able to enlighten you on how his BG7TBL clock helps in his system.

 

Below is a an image of one of those very 10Mhz clocks using a  75 Ohm 0.5m bnc lead being measured at 50Ohm (using the 50 ohm output on the clock)  you can clearly see the impudence mismatch and the resultant distorted square wave form.

 

Followed by a eye diagram again clearly visible is the cable mismatch finally the FFT plot of the resultant harmonic noise.

 

These clocks offer a respectable upgrade for your ether regen I would suggest the use of a ultra low noise power supply is essential with this device to bring out its potential.

 

 

 

 

 

 

 Would these https://www.pasternack.com/50-ohm-bnc-male-75-ohm-bnc-female-matching-pad-pe7006-p.aspx be appropriate to match the 50 ohm output to a 75 ohm cable & receiver?

[Clockmeister]

Hello AbAndgate

 

Old skool boolean logic been a while since collage for that one!

 

A great many DIY electronics enthusiasts tend to use  these impedance converting devices, its not really a complete solution if you are looking for a true impedance match.  Ideally all things being equal keeping the pathway the same Z all the wat through from Tx to Rx.

 

I appreciate however that most of the after-market clocks are specific 50 Ohms.

When we measure a 75 ohm impedance path ways and cables we use a set of true 75 Ohm calibration tools and have our VNA or TDR devices set to 75Ohm. Totally understand 99.7% of audio users are not going to have this facility.

 

What I can do is set up an example of a 75 ohm cable being used with 50 Ohm measurement parameters, this will give you an understanding of how the reflections can be seen on the waveform.

 

Think of it like a converter from SPDIF<>AES or single end to balanced always something lost in translation I feel.

 

Will it be better than a total mis match 50<>75? possibly yes, possibly no, sorry for not being more definitive.

 

 

[LowMidHigh]

4 hours ago, Clockmeister said:

Hello AbAndgate

 

Old skool boolean logic been a while since collage for that one!

 

A great many DIY electronics enthusiasts tend to use  these impedance converting devices, its not really a complete solution if you are looking for a true impedance match.  Ideally all things being equal keeping the pathway the same Z all the wat through from Tx to Rx.

 

I appreciate however that most of the after-market clocks are specific 50 Ohms.

When we measure a 75 ohm impedance path ways and cables we use a set of true 75 Ohm calibration tools and have our VNA or TDR devices set to 75Ohm. Totally understand 99.7% of audio users are not going to have this facility.

 

What I can do is set up an example of a 75 ohm cable being used with 50 Ohm measurement parameters, this will give you an understanding of how the reflections can be seen on the waveform.

 

Think of it like a converter from SPDIF<>AES or single end to balanced always something lost in translation I feel.

 

Will it be better than a total mis match 50<>75? possibly yes, possibly no, sorry for not being more definitive.

 

 

 

Logic suggests that converting impedences would outperform a 50/75 mismatch. But only people with your type of equipment and knowledge can test test that hypothesis. 

 

On a different note of yours:

"Think of it like a converter from SPDIF<>AES or single end to balanced always something lost in translation I feel." 

You feel, or that's a fact? Again, logic would poise said loss, but can you validated with measurements? 

 

 

[Clockmeister]

Hello LMH

 

Logic is a great for starting the process of investigation and exploration of an interesting subject.

 

Personally I not use any adapters in my audio system, from experience to much of a sonic compromise for myself.

 

In the lab unless a particular experiment calls for specific set of conditions that requires  adpters then they are not used

 

This evening I can perform those tests with the impedance cables and adapters

 

However with your Mutec you have the ability to try some basic tests for your self, the ref 10 has both 50 and 75 Ohm outputs

 

BNC outputs 1 & 2 are dedicated 75 Ohm and 3 to 8 are 50 Ohm assuming you are using a correctly terminated 75 Ohm cable for your connection between ref 10 and mc3

 

You could engage a 50 ohm output and swap the output cable  to mc3 and report back and see if you feel there is difference in the sound quality

 

 

 

 

 

 

[Clockmeister]

As promised some results from using impedance mis matched cables and adapters.

 

I used the lab's Mutec Ref10 SE-120 master clock as the signal generator, scope has been calibrated, all test cables 18Ghz reference ultra low loss 50 ohm, audio cables were 75 ohms matched. 50 Ohm adapter is an R&S reference 8Ghz model.

 

First images are a base line set, 50 ohm output, 50 ohm feed 50 ohm receiver, also I enhanced the waveform to highlight the very slight overshoot to give you an idea of  of a close up of the wave form top edge definition.

 

 

[Clockmeister]

Next set of images show a 50 ohm cable being fed a true 75 ohm signal to a 50 ohm receiver note the reduced PK<>PK voltage and difference in the top edge wave form

[Clockmeister]

Now a genuine 75 ohm audio clock cable using the 50 feed into a 50 ohm receiver plus the top edge definition again

[Clockmeister]

Last up 75ohm cable, 75 ohm feed i to a 50 Ohm receiver, the lastly  adapter added in to the chain for comparison.

 

As you can see the degradation of the wave form is pretty obvious, and especially with word clocks the wave form actually looks like a square wave form with very little deviations is the more ideal the situtaion

[LowMidHigh]

1 hour ago, Clockmeister said:

As promised some results from using impedance mis matched cables and adapters.

 

I used the lab's Mutec Ref10 SE-120 master clock as the signal generator, scope has been calibrated, all test cables 18Ghz reference ultra low loss 50 ohm, audio cables were 75 ohms matched. 50 Ohm adapter is an R&S reference 8Ghz model.

 

First images are a base line set, 50 ohm output, 50 ohm feed 50 ohm receiver, also I enhanced the waveform to highlight the very slight overshoot to give you an idea of  of a close up of the wave form top edge definition.

 

 

 

The photos are explanation are imminently clear. Thank you!

 

So in a case of mismatch, what would be most detrimental to the operation of the enslaved device?

 

1) Would the signal occasionally not trigger gate? It doesn’t seem the case.

2) Would the timing be off? The raise and fall in the mismatch cases seem slanted; not as plumb as when impedances are matched.

3) Would the ripple/reflections introduce fresh jitter?

3) Something else?

[Clockmeister]

LMH

 

For the incoming data stream not to trigger either the wave form would have to be very badly distorted on the leading edge or the overshoot would need to be in the high double figures.  another possible option could be the clock recovery circuit being of mediocre design with fairly high corner stone frequency and very average PLL ability this would lead to issues and a large increase in distorted sounds if not drop outs so pretty rare for this situation to happen

 

Yes the alignment of the audio data to word bit data would not be ideal at all, simply as the main timing (derived from the word clocking circuitry) would be sub optimal and therefore an increase in time interval error (TIE) would occur with out question

 

Any wave form deviation from the normal incoming wave form that point the circuit would usually see during normal operations can result in an in jitter.

 

These deviations are cause by the following:-

 

TRANSMISSION LINE problems, impedance mis matching all the whole line no just the cables or board connectors, circuit board embedded pathways, components in that circuit where inductorsm capacitors or resistors.

 

BANDWIDTH data line not being sufficiently capable of handling the full amount of transmitted data remember audio word clocks are a staple 10Mhz so a pretty low level data stream required, however if all of the those components that make that pathway are not up to the task then jitter will occur. We use a tool called an eye diagram to look at serial data transmission lanes and can determine by observing the decoded patterns where a great many issues actually line in the circuit.

 

CROSSTALK  pretty obvious where other signals which should be totally isolated from the clocking sections of device actually interferes with the clock for data circuit that is connected to the clock causing introduced noise which leads to unwanted induced jitter

 

 

Data generated jitter which is caused by bad programming at the core cpu or PFGA etc this can generate ISI (intersymbal interference) induced jitter and duty cycle distortion this is quite common in either highly complex devices where a lot of processing power is required to run a device and it takes a long time for total debug or in a simple device with novice programmer skills

 

We have special seial data analysis tools to look at all these types of jitter problems more  on that later

 

RANDOM NOISE

 

which is cause by thermal flunctations the higher the temperature the more jitter.

 

This next one can be difficult to understand ,shot noise which is the random movement of electrons within the circuit like black holes and large celestial bodies mbe moved by dark matter not a lot you can do to take into account for this issue!

 

Lastly in this section frequency noise or pink noise the lower the frequency the more noise is introduced into the design all if these items designs take into account when coming up with a new product so tradeoffs are going to happen

 

All of the above cited problems can cause data line data corruption problems which will lead to the Rx end of your data stream incorrectly reading the edge crossing if the data which will result in jitter factor of electronics life.

 

There are more items that need addressing with jitterwe have  just touched on the surface however that is for a later post.

 

 

[AnAndGate]

21 hours ago, Clockmeister said:

As promised some results from using impedance mis matched cables and adapters.

 

I used the lab's Mutec Ref10 SE-120 master clock as the signal generator, scope has been calibrated, all test cables 18Ghz reference ultra low loss 50 ohm, audio cables were 75 ohms matched. 50 Ohm adapter is an R&S reference 8Ghz model.

 

First images are a base line set, 50 ohm output, 50 ohm feed 50 ohm receiver, also I enhanced the waveform to highlight the very slight overshoot to give you an idea of  of a close up of the wave form top edge definition.

 

 

Thanks  so much for the testing.

So as I understand it you did 

1. 50 ohm Source > 50ohm cable > 50ohm receiver - the control
2. 75 ohm Source > 50ohm cable > 50ohm receiver - mismatched at 1 end: looked pretty good to me ?
3. 50 ohm Source > 75 ohm cable > 50ohm receiver - mismatched at both ends
4. 75 ohm Source > 75 ohm cable > 75to50ohm adapter > 50ohm receiver 

 

 

Am I interpreting the results correctly?  4 is better than 3. But worse than 2.

Implying adapter may make things worse . 

 

But we changed 2 variables between 2 and  4. 

To show if the adapter makes things worse, would this be valid or you feel it is the same as 2. 

       5. 75 ohm Source > 75 ohm cable > 50 ohm receiver

 

—

For the kit referred to earlier in this thread the ideal test would be 

 

50 ohm Source > 50 to 75 ohm adapter > 75 ohm cable > 75 ohm receiver 

Compared to 

50 ohm Source > 75 ohm cable > 75 ohm receiver 

 

(Mocking 50 ohm ref clk and 75 ohm ether region receiver)

 

I guess test 4 did the opposite 75ohm Source -> 50 ohm receiver so can we assume  the principle / reflections would be the same ?

 

I was so wanting the adapters to make a positive difference!!

[AnAndGate]

3 minutes ago, AnAndGate said:

Thanks  so much for the testing.

So as I understand it you did 

1. 50 ohm Source > 50ohm cable > 50ohm receiver - the control
2. 75 ohm Source > 50ohm cable > 50ohm receiver - mismatched at 1 end: looked pretty good to me ?
3. 50 ohm Source > 75 ohm cable > 50ohm receiver - mismatched at both ends
4. 75 ohm Source > 75 ohm cable > 75to50ohm adapter > 50ohm receiver 

Thanks so much for that 

 

Am I interpreting the results correctly?  4 is better than 3. But worse than 2.

Implying adapter may make things worse . 

 

But we changed 2 variables between 2 and  4. 

To show if the adapter makes things worse, would this be valid or you feel it is the same as 2. 

       5. 75 ohm Source > 75 ohm cable > 50 ohm receiver

 

—

For the kit referred to earlier in this thread the ideal test would be 

 

50 ohm Source > 50 to 75 ohm adapter > 75 ohm cable > 75 ohm receiver 

Compared to 

50 ohm Source > 75 ohm cable > 75 ohm receiver 

 

(Mocking 50 ohm ref clk and 75 ohm ether region receiver)

 

I guess test 4 did the opposite 75ohm Source -> 50 ohm receiver so can we assume  the principle / reflections would be the same ?

 

I was so wanting the adapters to make a positive difference!!

re reading your post, I think your last post covers with and without adapter is that right? 

 

[Clockmeister]

Good Morning AnAndGate

 

Thinking about at the above posts I should have constructed a finial post to condense this information into bullet points.

 

Ideal conditions:-

 

Zero compromise:-

 

Transmission 50 Ohm output <> 50 Ohm Cable <> 50 Ohm receive end terminal

 

Transmission 75 Ohm output <> 75 Ohm Cable <> 75 Ohm receive end terminal

 

Not so good:-

 

Transmission 50 Ohm output <> 75 Ohm Cable <> 50 Ohm receive end terminal

 

Transmission 75 Ohm output <> 50 Ohm Cable <> 75 Ohm receive end terminal

 

Transmission 75 Ohm output <> 75 Ohm Cable <> 50 Ohm receive end terminal

 

Transmission 50 Ohm output <> 75 Ohm Cable <> 75 Ohm receive end terminal

 

Least desirable:-

 

Transmission 50 Ohm output <> 75 Ohm Cable <> 50 Ohm <>  75 Ohm adapter at receive end terminal

 

Transmission 75 Ohm output <> 75 Ohm Cable <> 75 Ohm <>  50 Ohm adapter at receive end terminal

 

What you are attempting to achieve is removing as many of the reflection pathways as possible in the same way in your audio system you are trying to remove as much, electrical, mechanical, vibrational and radio reference noise as possible.

 

You make a comment on one of the results where the using a 75 Ohm cable on the 50 Ohm Tx & Rx  positions resulted in a not too dissimilar wave form than the ideal impedance matched example.

 

Later on I will delve onto some jitter analysis with eye diagrams to illustrate that even small amounts of deviation will not yield the  desired outcome. Also show how an ok looking square wave form can in fact had a multitude of sins.

 

Having spent a lot of years with this subject and many digital audio projects I can safely and happily share my thoughts, you can quite easily determine an impedance mismatch digital audio cable against the correctly matched item.

 

With clock cables this can manifest itself as less than ideal sound staging,  more diffuse vocal  placement, the timing is less coherent with an overall sense if it's not quite correct.

 

For those of you that use a a word clock and have genuine impedance matched cables, just try changing it for say a normal analogue cable of the same termination or a 50Ohm lab cable. Let us know what you feel.

 

Sorry back to the subject of adapters an issue personally I will not compromise on in any form in my professional and audio life.

 

Yesterday I received a large delivery of supplies for a new project I am working on and along with the components with another small box inside was pair of matching 50<>75 balun's BNC terminated for another project which has been on the back burner due to myself needing to construct a test rig for characterizing insertion loss for a/c mains common mode filters.

 

This means I have how some rather decent 50<>75 Ohm isolation transformers so this may work out to an acceptable compromise for some individuals.

 

These are NOT audio products but genuine lab test standard isolation transformers with impedance transitions by a company called North Hills if I have time today I will test it for you.

 

Below is wave form analysis of a Mutec MC3+ in word clock output mode using 176.4Khz files this will give you an idea of  how some smaller companies and individuals actually go into deeper investigations with there research and designing stages for their products I suspect Uptone have this philosophy as well.

 

[AnAndGate]

On 6/9/2020 at 11:26 PM, MartinT said:

 

Thanks - I will try it once I've got the impedance sorted. I hope I'm right in thinking an impedance mismatch affects a squarewave worse than a sinewave.

Hi @MartinT,

I am trying to understand Mutec mc-3 clock distribution capabilities:  given both Mutec and Etherregen may prefer the square wave signal  (and there is only one output) from the BG7TBL OCXO. Can Mutec MC-3 be fed the square wave, and act as the distributor through one of the word clk outputs to the Etherregen; as well as using the reference clock for  USB reclocking?

 

thanks

[Clockmeister]

AnAndGate

 

Mutec's MC3 is a word clock generator with outputs that mirror the incoming audio data sample rate it automatically changes rates according to the file sample rates

This uses the pcm reclocking feature to generate the outgoing word clock rate.

 

I believe  Martin T uses his Aliexpress sourced 10Mhz clock to slave both the Ether regen and Mutec MC3 with a master 10 Mhz signal directly from that clock.

 

The Mutec can be used as clocking distribution hub however the sample rates are derived from the incoming pcm data via one of its five different format inputs.

 

Brands such as Esoteric and Dcs can utilise the word clock output sample rates by using the outgoing extracted word clock go their individual WC inputs

 

Dcs have the added complexity of having two word clock sample rates  one at direct multiples of 44,1khz the other at multiples of 48Khz.

 

Each word clock lane automatically detects its the desired sample rate and selects the appropriate clock lane for the correct  SR multiple for the required file based playback  or incoming subscription service stream rate

[LowMidHigh]

37 minutes ago, AnAndGate said:

Hi @MartinT,

I am trying to understand Mutec mc-3 clock distribution capabilities:  given both Mutec and Etherregen may prefer the square wave signal  (and there is only one output) from the BG7TBL OCXO. Can Mutec MC-3 be fed the square wave, and act as the distributor through one of the word clk outputs to the Etherregen; as well as using the reference clock for  USB reclocking?

 

thanks

Short answer, no. 

 

I believe you can get that clock custom made with 2 square wave outputs, both 75ohm. There was a discussion about it in this thread. 

[MartinT]

Agreed with the previous responses, you cannot use the MC-3 to redistribute the 10MHz clock.

 

My new OCXO with 75 ohm impedance and all-squarewave outputs is on its way so I shall see whether it was worth it soon.

[AnAndGate]

Thanks all for clarifying. 

 

 

[kelvinwsy]

Here is myImpedance matching adapter😅

50ohm out of clock and 75 ohm end into ER

Still burning in but sounds promising

[AnAndGate]

1 hour ago, kelvinwsy said:

Here is myImpedance matching adapter😅

50ohm out of clock and 75 ohm end into ER

Still burning in but sounds promising

I bet you need to put a weight on the clock to keep it from keeling over! Doing something very similar...

My BG7TBL clock arrived this week. Internal OCXO spec is not as good as described earlier in the thread. I have the DAPU O55F-HSDD vs DAPU O55A OCXO.

I got some 50 -75 ohm matching pads that I attach to the clk out, and 75 ohm cables to etherregen & mutec mc-3. All seems to work. No kit to measure how much better (if any) the actual signal is with the pads, but will experiment with my ears. I notice on the cybershaft website they sell the 50 -75 ohm matchers too at a very modest price compared to non-audiophile matchers. Never usually see the pricing this way round.

 

 

 

[kelvinwsy]

Yeh 2200YEN but my adapter is so much more Sexy....

[kelvinwsy]

Oh no it is bidirectional .. 75ohm to 50 ohm csn work too! Just that I sm using 50 ohm out of the  clock into the ER 

[sgr]

A friend told me you could request the merchant at  (Ali baba) to have this device configured as a 50 ohm or 75 ohm and then you wouldn’t have to use the converter. 
 

I’ve never ordered from Ali Baba before and couldn’t ever complete my order. So I don’t know for sure. Compared Tuesday amazon their ordering process seems a bit sketchy. 
 

Does anyone sell these on eBay? 

Dies anyone know if a good linear power supply for it? 
 

thanks,